How Artificial Intelligence and Deep Learning apply in Ophthalmology
In Data Labeling in a Nutshell, I’ve mentioned that the application of artificial intelligence, ML and DL into life has brought specific positive changes. Widely applied in speech recognition and natural language processing, Artificial intelligence and Deep Learning have only begun to be used in healthcare in recent years. Utilizing computer algorithms to analyze detailed medical data to find relationships and produce diagnostic support outputs is artificial intelligence (AI) in healthcare.
AI and DL in ophthalmology have been used to analyze fundus images, optical coherence tomography, and visual fields, achieving robust classification performance in the identification of glaucoma-like disc, diabetic retinopathy (RD), retinopathy of prematurity (ROP), macular edema, and age-related macular degeneration (ARMD).
Let’s get into detail!
Artificial Intelligence and Deep Learning in Ophthalmology
AI and DL have found important roles in ophthalmology in early screening and detection of conditions such as:
- Diabetic retinopathy (DR)
- Age-related macular degeneration (ARMD)
- Retinopathy of prematurity (ROP)
- Glaucoma
To AI learn the ground truth, it had to be fed by medical datasets by using this Top 5 Free Medical Image Labeling Tool for Machine Learning in 2023
Diabetic Retinopathy (DR)
About DR
Diabetic retinopathy is an eye disease caused by diabetes. People with diabetes can have an eye disease called diabetic retinopathy. This is when high blood sugar levels cause damage to blood vessels in the retina.
AI application
In some published investigations, DL algorithms were used to diagnose microaneurysms, hemorrhages, hard exudates, cotton-wool patches, and neovascularization in DR patients. Some of these algorithms layer k-nearest neighbor (KNN), lattice neural network with dendritic processing (LNNDP), and morphological component analysis (MCA) on top of ANNs (kNN).
Gulshan and colleagues 1 from Google AI Healthcare reported a DL system with an excellent diagnostic performance which was developed using 128 175 retinal images to train model ai, graded between 3 and 7 times for DR and DMO by a panel of 54 US-licensed ophthalmologists and ophthalmology residents between May and December 2015.
A minimum of seven US board-certified ophthalmologists evaluated the test set, which included roughly 10,000 photos taken from two publicly accessible databases (EyePACS-1 and Messidor-2), with high intragrader consistency.
Age-related macular degeneration (ARMD)
About ARMD
ARMD is an eye disease that can cause blurry center vision and it happens when aging causes damage to the macula the part of the eye that controls sharp, straight-ahead vision
AI application
Algorithms of AI, ML, and DL provide automated detection of drusen, fluid, and geographic atrophy concerning AMD lesions to improve AMD diagnosis and treatment by using fundus images and spectral-domain OCT (SD-OCT). The automatic drusen detection with AI, ML and DL is likely to help ophthalmologists to improve the early and rapid diagnostic performance of fundus images
Burlina et al developed an automated grading for detecting AMD from color fundus images by using DL methods and AI, namely, DCNN. They claimed that a DL-based automated assessment of AMD was consistent with manual professional evaluation and that automated algorithms could play a critical role in the present management of AMD, address costs of screening or monitoring, access to healthcare, and the appraisal of novel treatments.
Ting et al 2 reported a clinically acceptable DL system diagnostic performance in detecting referable AMD. Specifically, the DL system was trained and tested using 108 558 retinal images from 38 189 patients
Retinol of prematurity (ROP)
About ROP
ROP is an eye disease that can happen in babies who are premature (born early) — or who weigh less than 3 pounds at birth. ROP happens when abnormal blood vessels grow in the retina (the light-sensitive layer of tissue in the back of your eye).
AI application
There are two main obstacles to the efficient application of ROP screening: ROP diagnosis is subjective, with significant interexaminer variability in the diagnosis leading to inconsistent application of evidence-based interventions; and there are insufficiently trained examiners in many parts of the world.
Early attempts to apply DL for automated ROP diagnosis have been made,19 66 which may be able to overcome both implementation difficulties for ROP screening. Most recently, Brown Et Al 3 published the findings of a completely automated DL system that could identify plus disease, the most significant aspect of severe ROP, with an AUC of 0.98 compared to a consensus reference standard diagnosis combining image-based diagnosis and ophthalmoscopy.
Glaucoma
About Glaucoma
Glaucoma is a group of eye conditions that damage the optic nerve.
Computer algorithms were trained by Li et al 4 and Ting et al 5 to recognize the glaucoma-like disc, which is indicated by a vertical CDR of 0.7 and 0.8, respectively.
AI application
The results of these tests include a summary of global indices, age-matched normative comparisons, and reliability factors. To assess the visual field.
Elze et al. 49 created an unsupervised computer algorithm that identifies clinically pertinent visual field loss patterns and gives each one a weighting coefficient. This technique is effective in identifying early visual field loss from glaucoma.
Archetype analysis with 16 visual fields (Visual Field) archetypes (ATs) that were derived from an unsupervised computer algorithm described by Elze et al.
Reference
Ophthalmology, D.of, Promising Artificial Intelligence-machine learning-deep…: The asia-pacific journal of ophthalmology, LWW. Available at: https://journals.lww.com/apjoo/Fulltext/2019/05000/Promising_Artificial_Intelligence_Machine.13.aspx (Accessed: January 10, 2023).
Ting, D.S.W. et al. (2019) Artificial Intelligence and deep learning in ophthalmology, The British journal of ophthalmology. U.S. National Library of Medicine. Available at: https://www.ncbi.nlm.nih.gov/pmc/artihttps://www.ncbi.nlm.nih.gov/pmc/articles/PMC6362807/cles/PMC6362807/ (Accessed: January 10, 2023).
Thanks for reading!
If you are finding information about machine learning, artificial intelligent or data in general or medical field. Follow us to acquire more useful knowledge about this 3 keywords.
Contact
Email: info@vinlab.io
Twitter: https://twitter.com/VinLab6
YouTube: https://www.youtube.com/@Vinlab-MedicalImageAnnotation
Open source project: https://github.com/vinbigdata-medical/vindr-lab
